Desloratadine alleviates ALS-like pathology in hSOD1^G93A mice via targeting 5HTR_2A on activated spinal astrocytes

Abstract

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease with progressive loss of motor neurons in the spinal cord, cerebral cortex and brain stem. ALS is characterized by gradual muscle atrophy and dyskinesia. The limited knowledge on the pathology of ALS has impeded the development of therapeutics for the disease. Previous studies have shown that autophagy and astrocyte-mediated neuroinflammation are involved in the pathogenesis of ALS, while 5HTR_2A participates in the early stage of astrocyte activation, and 5HTR_2A antagonism may suppress astrocyte activation. In this study, we evaluated the therapeutic effects of desloratadine (DLT), a selective 5HTR_2A antagonist, in human SOD1^G93A (hSOD1^G93A) ALS model mice, and elucidated the underlying mechanisms. HSOD1^G93A mice were administered DLT (20 mg·kg⁻¹·d⁻¹, i.g.) from the age of 8 weeks for 10 weeks or until death. ALS onset time and lifespan were determined using rotarod and righting reflex tests, respectively. We found that astrocyte activation accompanying with serotonin receptor 2 A (5HTR_2A) upregulation in the spinal cord was tightly associated with ALS-like pathology, which was effectively attenuated by DLT administration. We showed that DLT administration significantly delayed ALS symptom onset time, prolonged lifespan and ameliorated movement disorders, gastrocnemius injury and spinal motor neuronal loss in hSOD1^G93A mice. Spinal cord-specific knockdown of 5HTR_2A by intrathecal injection of adeno-associated virus9 (AAV9)-si-5Htr2a also ameliorated ALS pathology in hSOD1^G93A mice, and occluded the therapeutic effects of DLT administration. Furthermore, we demonstrated that DLT administration promoted autophagy to reduce mutant hSOD1 levels through 5HTR_2A/cAMP/AMPK pathway, suppressed oxidative stress through 5HTR_2A/cAMP/AMPK/Nrf2-HO-1/NQO-1 pathway, and inhibited astrocyte neuroinflammation through 5HTR_2A/cAMP/AMPK/NF-κB/NLRP3 pathway in the spinal cord of hSOD1^G93A mice. In summary, 5HTR_2A antagonism shows promise as a therapeutic strategy for ALS, highlighting the potential of DLT in the treatment of the disease.

DLT as a 5HTR_2A antagonist effectively promoted autophagy to reduce mutant hSOD1 level through 5HTR_2A/cAMP/AMPK pathway, suppressed oxidative stress through 5HTR_2A/cAMP/AMPK/Nrf2-HO-1/NQO-1 pathway, and inhibited astrocytic neuroinflammation through 5HTR_2A/cAMP/AMPK/NF-κB/NLRP3 pathway in the spinal cord of hSOD1^G93A mice.

Introduction

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by the progressive loss of motor neurons in the spinal cord, cerebral cortex and brain stem [1]. The main clinical manifestation of this disease involves progressive muscle atrophy and dyskinesia, ultimately leading to respiratory failure and patient death [2]. The prevalence rate of ALS is approximately 2–3 per 100,000 persons, and approximately half of ALS patients die within three to four years after symptom onset [3]. To date, the Food and Drug Administration (FDA) has only approved two anti-ALS drugs, the glutamate receptor antagonist Riluzole [4] and free radical scavenger edaravone [5], although these two reagents only slightly relieve the pathological symptoms of ALS rather than heal or cure this disease [6]. Currently, ALS is a serious physical, medical, and economic burden to individuals and society.

The aggregation of misfolded proteins is an important pathological feature of neurodegenerative diseases, and some toxic protein aggregates have been identified in ALS pathology [7], including tar DNA binding protein 43 (TDP-43) [8] and SOD1 [9] protein mutants. Notably, the G93A mutation protein in SOD1 (SOD1^G93A), which causes the abnormal precipitation of mutant SOD1 proteins, is present in 20%–25% of familial ALS cases [3, 10]. SOD1^G93A protein aggregation can directly or indirectly impair motor neurons by mediating neuronal oxidative stress and glial neuroinflammation [11]. As a result, promoting the clearance of the SOD1^G93A protein is believed to be a potential strategy for the treatment of ALS [12]. Accumulating evidence has determined that autophagy, as an important process for toxic protein degradation in cells, plays a key role in degrading the SOD1^G93A protein in ALS [13], thus making autophagy a promising potential target for anti-ALS drug discovery.

As indicated in published reports, SOD1^G93A protein-induced neuroinflammation is one of the earliest pathological events in ALS [14], and a large number of inflammatory astrocytes and microglia are activated in the spinal cord of ALS patients and model mice [14, 15]. Astrocytes express and recognize SOD1^G93A protein in the spinal cord, and SOD1^G93A protein accumulation induces astrocytic activation and neuroinflammation [16] followed by damage to the surrounding neurons through the release of a series of inflammatory cytokines, including IL-1β and TNF-α [16]. In particular, the NLRP3 inflammasome, which is a multiprotein complex composed of the innate immune receptor protein NLRP3, adapter protein ASC and proinflammatory protease caspase-1, potently functions as an effector of the inflammatory response [17], and its activation occurs after spinal astrocytes carrying the SOD1^G93A protein are activated [15]. It was reported that activated Caspase 1, as an effector element, cleaves pro-IL-1β into mature IL-1β, which is released by activated astrocytes [18]. Additionally, a series of damage signaling factors, including reactive oxygen species (ROS) and chemokines, released by damaged motor neurons further activate adjacent glial cells, resulting in an exacerbated neuroinflammatory response [18]. All evidence has thus addressed the key role of astrocyte-mediated neuroinflammation in ALS pathology.

5HT receptor 2A (5HTR_2A), a 5HT₂ receptor subtype, is widely expressed in the spinal cord and participates in various biological functions [19]. It has been reported that 5HTR_2A is implicated in the inflammatory pain response and is abnormally upregulated in spinal cord injury [20]. Moreover, treatment with the 5HTR_2A antagonist sarpogrelate inhibited formalin-induced inflammation in mice [21]. Interestingly, strong 5HTR_2A immune reactivity in activated astrocytes was found in the brain tissues from patients with central nervous system (CNS) diseases (e.g., cerebral infarction, hypertensive encephalopathy, Alzheimer’s disease and Huntington’s disease), while such reactivity was not observed in normal or nonactivated astrocytes [22]. These results suggested that 5HTR_2A participates in the early stage of astrocytic activation and that 5HTR_2A antagonism may suppress astrocytic activation.

Herein, we determined that antiallergic drug desloratadine (DLT, Fig. 1a) [23], a specific antagonist of 5HTR_2A in our previously published work [24], could efficiently ameliorate ALS-like pathology in mice. The underlying mechanism was intensively investigated by assay against the ALS (hSOD1^G93A) model mice with 5HTR_2A knockdown selectively in the spinal cord by injection of adeno-associated virus AAV9-si-5Htr2a. Our findings have supported that 5HTR_2A antagonism shows promise as a therapeutic strategy for ALS and highlighted the potential of DLT in the treatment of this disease.

Fig. 1. 5HTR_2A was upregulated in the spinal astrocytes of hSOD1^G93A mice and ALS patients.

a Chemical structure of DLT. b RT-PCR results indicated that 5Htr2a was the most dysregulated 5HT receptor subtype among the tested 12 sub-types of 5HT receptors in the spinal cord of ALS mice. c The GEO database (GSE26927) results revealed that the mRNA level of 5HTR_2A was increased in ALS patients. d Western blot assay and (e) its quantification results indicated that the protein level of spinal 5HTR_2A in ALS mice was higher than that of WT mice. f Immunofluorescence assay and (g) its quantification results demonstrated that spinal astrocytic 5HTR_2A was upregulated in ALS mice. Scale bar: 10 µm. h Immunofluorescence assay and (i) its quantification results indicated that there was obvious spinal astrocytic activation in ALS mice. Scale bar: 80 µm. All values were presented as mean ± SEM. *P < 0.05, ***P < 0.001 compared by t test.

Materials and methods

Study design

The goals of the study were to evaluate the beneficial effects of the 5HTR_2A antagonist DLT on the ALS pathology of hSOD1^G93A mice (life cycle, dyskinesia, gastrocnemius injury, motor neuronal loss, neuronal oxidative stress and astrocytic inflammatory activation) and to investigate the underlying mechanism by establishing spinal cord 5HTR_2A knockdown in hSOD1^G93A mice with AAV9-si-5HTR_2A injection. For all animal studies, mice were litter-matched, age-matched and sex-matched to reduce the variance in the parameters of interest. Subsequently, histology, immunostaining, Western blotting and RT‒PCR assays of spinal cord tissues were performed. Cell-based assays against primary astrocytes or cell lines were carried out to verify the conclusions of the animal experiments.

Animal experiments

All animal experiments were conducted in strict accordance with the institutional ethical guidelines on animal care at Nanjing University of Chinese Medicine. All specific pathogen-free (SPF) animals were fed sterilized SPF pellet rodent feed and sterilized water and maintained under standard conditions with a room temperature of 22 °C and a 12 h light/dark cycle.

hSOD1^G93A (B6SJL-BTG, 1Gur/J, hSOD1^G93A) ALS model mice were purchased from Jackson Laboratory (USA) and carried a multicopy human SOD1 mutant (mutated from glycine at position 93 of the human SOD1 gene to alanine) [25]. hSOD1^G93A heterozygous male mice were crossed with wild-type (WT)-negative female mice (1:2). hSOD1^G93A transgenic (male and female) mice were selected from the offspring mice based on DNA electrophoresis analysis, and their littermate (male and female) WT mice were used as a negative control.

Experimental mice were divided into groups of WT mice treated with vehicle group (WT, male and female, n = 24 per group), WT mice treated with DLT-20 mg·kg^-1·d^-1 group (WT + DLT, male and female, n = 24 per group), hSOD1^G93A mice treated with vehicle (ALS, male and female, n = 24 per group), hSOD1^G93A mice treated with DLT-20 mg·kg^-1·d^-1 (ALS + DLT; male and female, n = 24 per group), WT mice injected with AAV9-si-NC and treated with vehicle (WT-NC; male and female, n = 24 per group), hSOD1^G93A mice injected with AAV9-si-NC and treated with vehicle (ALS-NC; male and female, n = 24 per group), hSOD1^G93A mice injected with AAV9-si-5Htr2a and treated with vehicle (ALS-KD; male and female, n = 24 per group) and hSOD1^G93A mice injected with AAV9-si-5Htr2a and treated with DLT-20 mg·kg^-1·d^-1 (ALS-KD + DLT; male and female, n = 24 per group).

DLT was dissolved in vehicle (normal saline containing 6% Tween 80) and administered by gavage [24]. Experimental mice were administered vehicle or DLT at the age of 8 weeks. Some of the experimental mice (n = 12 per group) were sacrificed at the age of 18 weeks, and blood, muscle and spinal cord tissues were collected for subsequent pathological, biochemical, and immunological analyses. The remaining experimental mice (n = 12 per group) were continuously treated with vehicle or DLT until death to determine the onset time and survival rate of hSOD1^G93A mice. hSOD1^G93A mice with spinal cord-specific 5HTR_2A knockdown were prepared by treatment with AAV9-si-5Htr2a (1 × 10¹¹ vg/ mouse) through intrathecal injection at 7 weeks of age. The knockdown efficiency of AAV9-si-5Htr2a in the spinal cords of ALS model mice (motor neurons were labeled by SMI-32; astrocytes were labeled by GFAP) was confirmed by Western blot and immunofluorescence assays (Fig. S1a–d).

Experiment mice at the age of 8 weeks were subjected to weekly behavioral experiments including the rod rotation test, gait analysis and grip strength test.

Body weight monitoring

Given the significant weight loss in the ALS model mice after disease onset, the body weight of the mice was measured weekly by an electronic scale from the age of 8 weeks. To exclude the influence of diurnal change on the body weight of mice, weight measurement was routinely performed at 9:00 am.

ALS onset time and survival rate

The ALS onset time of hSOD1^G93A mice was defined as the time when the mouse first dropped from the rotarod within 180 s, and the death date of hSOD1^G93A mice was defined as the time when the mouse lying on its side failed to complete the righting reflex within 30 s [13].

Rotating rod test

The fore/hind-limb motor coordination and balance function of mice were evaluated by measuring their time on the rotating rod [26]. Briefly, the experiment mice were subjected to a rotating rod test, which was performed by using a rotarod turn bar meter (12 RPM). Each mouse was tested three times per training, and the longest staying time on the rotating rod was recorded.

Mouse gait analysis

The abnormal gait of hSOD1^G93A mice was observed by mouse gait analysis [27]. Briefly, the front and back feet of mice were coated with red and green nontoxic pigments, respectively. The mice were allowed to walk along a track (length: 50 cm; width: 10 cm), and gaits were recorded. The experiment was performed three times once a week. The distance between steps obtained from three independent experiments were averaged to determine mouse gait.

Hanging cage test

The grip strength of the mice was analyzed by the hanging cage test [28]. Briefly, the longest time for mice to hold the cage cover was recorded (the time limit was set as 90 s). The test was performed three times once a week, and the longest hanging time was recorded.

Mouse spinal cord tissue preparation

Mice were anesthetized with pentobarbital sodium (1%, 1 mL/kg, i.p.) and perfused with normal saline. The spinal cord of the mice was then removed and flushed out with normal saline.

For the Western blot analysis, a piece of spinal cord tissue (approximately 10 mg) was taken into a precooled 1.5 mL EP tube followed by the addition of 400 μL RIPA lysis buffer containing 1% protease inhibitors and 1% phosphorylase inhibitors, and the tissues were fully homogenized by an ultrasound instrument and lysed on ice for 30 min. After centrifugation at 12,000 rpm at 4 °C for 30 min, the supernatant was taken, and the protein concentration was measured by a BCA detection kit (Beyotime, China). Finally, the protein samples were mixed with 2× loading buffer at a ratio of 1:1 and stored at −20 °C.

For the qPCR assay, a piece of spinal cord tissue (approximately 10 mg) was placed into a precooled 1.5 mL EP tube followed by the addition of 300 μL TRIzol, and the tissues were ground into a nonvisible tissue block by using a tissue mill. Then, 700 μl TRIzol was added to each tube and lysed for 30 min on ice. The follow-up procedure was performed according to the manufacturer’s instructions (Takara Bio, China). The sequences of primers used in the experiments are listed as follows (5'-3').

Mouse Nrf2: forward, TTCTTTCAGCAGCATCCTCTCCAC

Mouse Nrf2: reverse, ACAGCCTTCAATAGTCCCGTCCAG

Mouse HO-1: forward, CAAGCCGAGAATGCTGAGTTCATG

Mouse HO-1: reverse, GCAAGGGATGATTTCCTGCCAG

Mouse NQO-1: forward, GGATTGGACCGAGC-TGGAA

Mouse NQO-1: reverse, AATTGCAGTGAA-GATGAAGGCAAC

Mouse GAPDH: forward, ACAGCAACAGGGTGGTGGAC

Mouse GAPDH: reverse, TTTGAGGGTGCAGCGAACTT

For immunofluorescence and immunohistochemistry assays, the expanded lumbar segment (L4-L5 segment of the spinal cord) of mice was cut and fixed in 4% paraformaldehyde (PFA) overnight and then transferred to 30% sucrose solution. The tissues were fixed with OCT embedding agent and serially sectioned at a thickness of 10 μm. Finally, the tissue sections were stored in tissue cryopreservation solution (20 mL glycerol, 30 mL ethylene glycol, 2 mL DMSO and 48 mL PBS) at −20 °C.

For the Nissl staining assay [29], the frozen sections were placed in Cresyl violet staining solution and incubated for 10 at 56 °C. The sections were then washed with deionized water for 5 s, differentiated in Nissl differentiation solution for 2–3 s and rapidly dehydrated in absolute ethanol. Finally, the sections were mounted and sealed with neutral gum. Motor neurons (diameter >20 μm) with obvious nuclei localized in the anterior horn of the spinal cord were identified and counted.

Mouse muscle tissue preparation

The experimental mice in each group were anesthetized with 10% chloral hydrate, and the hind-limb gastrocnemius tissues (diameter: 5 mm; length: approximately 1 cm) of the mice were obtained and fixed with yarrow gum. The muscle tissues were put into precooled isopentane (approximately −160 °C), and the mixture was stirred to evenly freeze the tissue. The tissues were then stored in a −80 °C refrigerator.

A hematoxylin-eosin (HE) staining assay [30] was performed by using a commercial kit (Beyotime, China). Briefly, the muscle sections were sequentially stained with hematoxylin (10 min), differentiation solution (30 s), eosin (30 s), 95% ethanol (2–3 s), 100% ethanol (1 min), xylene I solution (1 min) and xylene II solution (1 min) and sealed with neutral gum.

A modified Gomori Trichrome staining assay [30] was performed by using a commercial kit (Solarbio, China). Briefly, the muscle sections were sequentially stained with hematoxylin (10 min), Gomori solution (10–20 min), 90% ethanol (1 min), 100% ethanol (1 min), xylene I solution (1 min) and xylene II solution (1 min) and sealed with neutral gum.

A NADH-tetrazolium reductase (NADH-TR) staining assay [31] was performed by using a commercial kit (Solarbio, China). Briefly, muscle sections were incubated in 30 mL of a specific solution (0.05 mol/L Tris hydrochloric acid buffer containing 30 mg nitro blue tetrazolium and 24 mg β-nicotinamide-adenine dinucleotide, pH: 7.4) at 37 °C for 30 min. Muscle sections were then sequentially immersed in different concentrations of acetone (30% → 60% → 90% → 60% → 30%) and sealed with glycerin gel.

Cell culture

Neuroblastoma-spinal cord (NSC) 34 motor neuron hybrid (NSC34) cells [32] were cultured in DMEM high glucose cell medium supplemented with 10% FBS and 1% PS.

For constructing NSC34 cells stably expressing hSOD1^G93A protein (hSOD1^G93A-NSC34 cells), NSC34 cells were seeded in 12-well plates and cultured with DMEM high glucose cell medium supplemented with 10% FBS and 1% PS. A lentivirus infection mixture (infection booster, fresh serum-free medium and lentivirus) was obtained and used to culture NSC34 cells for 12 h. The culture medium containing virus was then replaced with DMEM containing 10% FBS and 1% PS without virus. After 72 h, the infection efficiency was observed (80%), and the hSOD1^G93A-NSC34 cells were screened and maintained by using DMEM containing 10% FBS, 1% PS and 1 μM puromycin.

Primary astrocytes from the spinal cord of mice were dissected from newborn C57BL/6 N mice [33]. In brief, the spinal cord was placed in a cell culture dish, and the vascular membrane was peeled away gently under a microscope. The spinal cord tissue was then shredded and digested with trypsin at 37 °C for 30 min. The astrocytes were seeded in culture flasks precoated with PDL and cultured in DMEM/F12 supplemented with 10% FBS and 1% PS. After 24 h, the cells were treated with 1 μM cytosine β-D-arabinofuranoside for 48 h to prevent the proliferation of other cells, including microglia and fibroblasts.

MDA assay

The cells were seeded in a 12-well plate at a density of 1 × 10⁵ cells/mL. After pretreatment with DLT for 12 h, the medium was removed and replaced with medium containing DLT or/and hydrogen peroxide (300 μM) for 8 h. The cell lysate was centrifuged at 12,000 × g for 10 min at 4 °C, and the protein concentration was quantified by using a BCA protein concentration assay kit (Beyotime, China). Finally, the MDA level of the cell samples was determined according to the kit instructions (Jian Cheng, China).

Superoxide anion level detection

The superoxide anion level of cells was measured by using a dihydroethidium superoxide anion fluorescent probe (Beyotime, China). Briefly, cells were washed once with PBS and incubated with dihydroethidium (5 μM) for 30 min at 37 °C. The superoxide anion level was then determined by using a fluorescence microscope (excitation wavelength at 300 nm, emission wavelength at 610 nm).

mTagRFP-mWasabi-LC3 plasmid transfection

The mTagRFP-mWasabi-LC3 plasmid [34] was transfected by using Lipofectamine 2000 transfection reagent. NSC-34 cells were seeded in cell culture plates and cultured overnight to a cell confluence of 50%. The transfection reagent (2 μL Lipofectamine 2000 and 1000 ng plasmid per 1 mL) was prepared with fresh OMEM medium without FBS. The cell medium was replaced with medium containing transfection reagent. After 8 h of transfection, the medium was replaced with new medium containing 10% FBS and 1% PS.

Statistical analysis

Kaplan‒Meier (K‒M) survival analysis was used to analyze the onset time and survival rate of mice. All experimental data are presented as the mean ± SEM. The significant difference between two groups of data was analyzed by Student’s t test, the significant difference between multiple groups of data was analyzed by one-way ANOVA, and P < 0.05 was considered a significant difference. Graphs were made by using GraphPad Prism 8.0.

For further details regarding the methods used in this study, please refer to the supplementary text.

Results

5HTR_2A was upregulated in the spinal astrocytes of hSOD1^G93A mice and ALS patients

5HTR_2A was upregulated in the spinal cord of hSOD1^G93A mice and ALS patients

Given that astrocyte-mediated neuroinflammation is one of the early pathologies of ALS and that 5HTR_2A plays a key role in inflammatory diseases [16, 22], we investigated whether 5HTR_2A dysregulation was present in hSOD1^G93A mice (hereinafter abbreviated as ALS model mice, if there is no special reference) and ALS patients.

RT‒PCR results (Fig. 1b) indicated that 5Htr2a was the most dysregulated 5HT receptor subtype among the 12 tested subtypes of 5HT receptors in the spinal cord of hSOD1^G93A mice (5Htr1e and 5Htr1f were not detected due to their rare expression in the spinal cord according to the search results, https://www.proteinatlas.org/, and the results also indicated the low level of 5Htr2b in the spinal cord). Additionally, the results (Fig. 1c) from the public GEO database (GSE26927) [35] (healthy people, n = 10; ALS patients, n = 10) also revealed that the mRNA level of 5HTR_2A was increased in ALS patients. Moreover, the Western blot results (Fig. 1d, e) demonstrated that the protein level of spinal 5HTR_2A was increased in ALS model mice.

5HTR_2A was upregulated in the spinal astrocytes of ALS model mice

Considering that 5HTR_2A is widely distributed in multiple cell populations of the spinal cord, the cell responsible for mediating 5HTR_2A upregulation was next identified by determining the colocalization of 5HTR_2A with motor neurons (SMI-32 [13]), astrocytes (GFAP [36]) and microglia (IBa1 [36]) by immunofluorescence assay.

The results indicated that 5HTR_2A was primarily colocalized with astrocytes compared with spinal motor neurons and microglia in both WT and ALS model mice (Fig. 1f, g). Moreover, consistent with published reports [1, 15], we also determined that there was obvious astrocytic activation in the spinal cords of ALS model mice (Fig. 1h, i). The results thus demonstrated that 5HTR_2A upregulation is closely associated with activated spinal astrocytes.

DLT delayed ALS onset time and improved lifespan in ALS model mice through 5HTR_2A

Given the above determination of the close association between 5HTR_2A upregulation and activated spinal astrocytes in ALS model mice, the 5HTR_2A antagonist DLT, which was identified in our previous work [24], was applied as a probe to investigate the role of 5HTR_2A regulation in the ALS-like pathology of mice.

The effects of DLT on ALS onset time and lifespan in ALS model mice were first assessed by rotarod [28] and righting reflex [27] tests. Additionally, considering that DLT as an antiallergic drug targets histamine receptor 1 (H1R) [37], the assays were also performed in ALS model mice with selective 5HTR_2A knockdown in the spinal cord by injection of AAV9-si-5Htr2a to verify 5HTR_2A, not H1R (or other targets), as the functional target for DLT in the current work (Fig. S1a, b). Figure 2a shows the experiment schedule.

Fig. 2. DLT delayed ALS onset time and ameliorated the dyskinesias of ALS mice through 5HTR_2A.

a Schedule of animal treatments and behavior tests. b, c Rotarod test results indicated that either DLT treatment or of AAV9-si-5Htr2a (ALS-KD) injection delayed the onset time of male ALS mice, and DLT treatment (DLT-20) had no impacts on ALS onset time in AAV9-si-5Htr2a injected male ALS mice (ALS-KD + DLT). d, e Righting reflex test results indicated that either DLT treatment or AAV9-si-5Htr2a injection (ALS-KD) prolonged the lifespan of male ALS mice, and DLT treatment (DLT-20) had no impacts on the survival time of AAV9-si-5Htr2a injected male ALS mice (ALS-KD + DLT). f, g Weight detection results indicated that either DLT treatment or AAV9-si-5Htr2a injection (ALS-KD) alleviated weight loss of male ALS mice, DLT treatment (DLT-20) had no impacts on weight of AAV9-si-5Htr2a injected male ALS mice (ALS-KD + DLT). h Schedule of animal treatments and behavior tests. i, j Rotarod test indicated that either DLT treatment or AAV9-si-5Htr2a injection (ALS-KD) prolonged the dropped time in male ALS mice, and DLT treatment (DLT-20) had no impacts on the dropped time in AAV9-si-5Htr2a injected male ALS mice (ALS-KD + DLT). k, l Gait analysis test results indicated that either DLT treatment or AAV9-si-5Htr2a injection alleviated the symptom of shortened stride length in male ALS mice (ALS-KD), and DLT treatment (DLT-20) had no impacts on the stride length in AAV9-si-5Htr2a injected male ALS mice (ALS-KD + DLT). m, n Hanging cage test results indicated that either DLT treatment or AAV9-si-5Htr2a injection (ALS-KD) prolonged the hanging time of male ALS mice and DLT treatment (DLT-20) had no impacts on the hanging time of AAV9-si-5Htr2a injected male ALS mice (ALS-KD + DLT). All values were presented as mean ± SEM. ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. *P < 0.05, ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

Rotarod test

The results (Fig. 2b for males; Fig. S2a for females) indicated that ALS onset time in mice appeared at the age of approximately 100 days (male ALS group = 97.6 ± 1.473 days; female ALS group = 107.2 ± 0.853 days), and DLT treatment delayed the ALS onset time in mice (male ALS + DLT = 103.4 ± 1.705 days; female ALS + DLT = 113.2 ± 0.953 days).

Righting reflex test

The results (Fig. 2d for male; Fig. S2c for female) indicated that the lifespan of ALS model mice was approximately 130 days (male ALS group = 126 ± 1.721 days; female ALS group = 130.4 ± 1.344 days), and DLT treatment prolonged the lifespan of ALS model mice (male ALS + DLT = 131.6 ± 1.51 days; female ALS + DLT = 135.9 ± 1.953 days). Additionally, the body weight monitoring results (Fig. 2f for males; Fig. S2e for females) also demonstrated that DLT treatment alleviated weight loss in ALS model mice.

Notably, DLT treatment had no impact on ALS onset time (Fig. 2b, S2a), lifespan (Fig. 2d, S2c) or weight (Fig. 2f, S2e) in WT mice. AAV9-si-5Htr2a injection delayed ALS onset time (Fig. 2c for male; Fig. S2b for female), prolonged lifespan (Fig. 2e for male; Fig. S2d for female) and alleviated weight loss (Fig. 2g for male; Fig. S2f for female) in ALS model mice, and DLT treatment had no impact on any of the abovementioned events in AAV9-si-5Htr2a-injected ALS model mice (Fig. 2c, e, g; Fig. S2b, d, f).

Thus, all results demonstrated that DLT delayed the ALS onset time and improved the lifespan of ALS model mice through 5HTR_2A.

DLT ameliorated dyskinesias of ALS mice through 5HTR_2A

Given that dyskinesia is one of the main clinical symptoms of ALS, relevant tests, including rotarod [28], gait analysis [27] and hanging cage [28] tests were carried out to evaluate the ability of DLT to ameliorate dyskinesia in ALS model mice. Figure 2h shows the schedule for the assays.

Rotarod test

The results (Fig. 2i for male; Fig. S2g for female) indicated that ALS model mice dropped from the rotarod around the onset time (male ALS group = 100 ± 1.33 days; female ALS group =114 ± 2.14 days), and DLT treatment prolonged the drop time from the rotarod (male ALS group = 110 ± 0.87 days; female ALS group =122 ± 1.84 days) of ALS model mice.

Gait analysis test

The results (Fig. 2k for male; Fig. S2i for female) indicated that the stride length of ALS model mice was shortened after ALS onset and that DLT treatment ameliorated such a mobility impairment in ALS model mice.

Hanging cage test

The results (Fig. 2m for male; Fig. S2k for female) indicated that the time for ALS model mice to drop from the cage cover was less than 90 s and close to the time of ALS onset, and DLT treatment prolonged the hanging time of ALS model mice. These results thus implied that DLT treatment improved muscle strength in ALS model mice.

Notably, DLT treatment had no impact on the movement functions (rotarod test, Fig. 2i and S2g; gait analysis test, Fig. 2k and S2i; hanging cage test, Fig. 2m and S2k) of WT mice. AAV9-si-5Htr2a injection improved dyskinesia as indicated in rotarod (Fig. 2j for male; Fig. S2h for female), gait analysis (Fig. 2l for male; Fig. S2j for female) and hanging cage (Fig. 2n for male; Fig. S2l for female) tests in ALS model mice, but DLT treatment had no impact on any of those tests in AAV9-si-5Htr2a-injected ALS model mice (Fig. 2j, l, n; Fig. S2h, j and l).

Thus, the results demonstrated that DLT treatment ameliorated the dyskinesia of ALS model mice through 5HTR_2A.

DLT ameliorated gastrocnemius injury in ALS model mice through 5HTR_2A

Given that gastrocnemius injury in ALS model mice can gradually lead to tremor paralysis, weakness, atrophy to final paralysis and further dyskinesias [31], we investigated the potential of DLT in ameliorating gastrocnemius injury in ALS model mice.

Considering that we had verified the alleviation of DLT on ALS-like pathology in both male and female mice and that the incidence of ALS in males is higher than that in females [3], the experiments in this section were only conducted with male mice to exclude the potential impact of gender difference on disease development.

DLT improved gastrocnemius pathomorphology

The HE staining results (Fig. S3a), showed the diameter reduction and disordered arrangement of muscle fibers and nuclear centralization in the gastrocnemius of ALS model mice, while DLT treatment ameliorated these muscle pathologies in ALS model mice.

DLT improved gastrocnemius mitochondrial dysfunction

Modified Gomori’s (Modified Gomori Trichrome, MGT) staining was used to investigate mitochondrial function in gastrocnemius fibers of mice [30]. The MGT staining results (Fig. S3b) showed obvious mitochondrial defects, including abnormal aggregation in the cytoplasm and a ragged-red fiber appearance caused by broken gastrocnemius fibers in the ALS model mice. DLT treatment improved the muscle mitochondrial dysfunction in ALS model mice.

DLT inhibited gastrocnemius oxidative stress

Oxidative stress (OS) is an important factor in gastrocnemius injury in ALS model mice [38], and type I muscle fibers are mainly associated with oxidative metabolism, while type II muscle fibers are mainly linked to glucose metabolism [31]. Here, a NADH-TR staining assay was performed, which stains type I and II muscle fibers darkly and lightly, respectively.

As indicated in Fig. S3c, DLT treatment reduced the number of darkly stained muscle fibers in the gastrocnemius, implying that DLT inhibited oxidative stress in the gastrocnemius of ALS model mice.

Notably, AAV9-si-5Htr2a injection in the spinal cord improved all of the abovementioned gastrocnemius dysfunctions (pathomorphology, Fig. S3a; mitochondrial dysfunction, Fig. S3b; oxidative stress, Fig. S3c) in ALS model mice, and DLT treatment exerted no effects on any of these gastrocnemius dysfunctions in AAV9-si-5Htr2a-injected ALS model mice (Fig. S3a–c).

To investigate the mechanism underlying the improvement of DLT on muscular dysfunctions, the protein level of 5HTR_2A in the muscle tissues of ALS model mice was detected by Western blot. As shown in Fig. S3d, e, no differences were observed in the muscle tissue 5HTR_2A levels between WT and ALS model mice. P-AMPK levels in the muscle tissues of ALS model mice were further detected due to the role of AMPK in muscle protection [39] and mitochondrial homeostasis [40]. The Western blot analysis results indicated that the protein level of p-AMPK in the muscle tissues of ALS model mice was reduced compared with that in WT mice, and DLT treatment increased the protein level of p-AMPK in the muscle tissues of ALS model mice (Fig. S3d, e). Interestingly, intrathecal injection of AAV9-si-5Htr2a had no effect on the protein level of 5HTR_2A in the muscle tissues of ALS model mice but increased the protein level of p-AMPK, and DLT treatment had no impact on p-AMPK in AAV9-si-5Htr2a-injected ALS model mice (Fig. S3d, f). Notably, public data from The Human Protein Atlas (https://www.proteinatlas.org/) showed that 5HTR_2A was highly expressed in the central nervous system but lowly expressed in muscle tissues. Thus, we speculated that the regulation of p-AMPK levels and muscular dysfunction by DLT was independent of its direct regulation of 5HTR_2A signaling in the muscle tissues of ALS model mice.

Therefore, the results demonstrated that DLT ameliorated gastrocnemius injury in ALS model mice through 5HTR_2A.

DLT protected against spinal motor neuronal damage in ALS model mice through 5HTR_2A

Considering that neuronal injury is the main cause of dyskinesia in ALS [41], we investigated the potential of DLT in protecting against spinal motor neuronal damage in ALS model mice.

DLT suppressed spinal motor neuronal loss in ALS model mice

The Nissl staining assay results indicated that the spinal motor neurons in the anterior horn of ALS model mice were lost and morphologically atrophied, and DLT treatment effectively reversed such motor neuronal loss in the spinal cords of ALS model mice (Fig. 3a, b).

Fig. 3. DLT protected against spinal motor neuronal damage in ALS mice through 5HTR_2A.

a Nissl staining assay and (b, c) its quantification results indicated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection reversed spinal neuron loss in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal neuron loss in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). Scale bar: 100 µm. d SMI-32 staining assay and (e, f) its quantification results indicated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection reversed spinal neuron loss in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal neuron loss in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). Scale bar: 30 µm. g TUNEL staining assay and (h, i) its quantification results indicated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection suppressed spinal neuron apoptosis in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal neuron apoptosis in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). Scale bar: 100 µm. All values were presented as mean ± SEM. ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. **P < 0.01, ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

Additionally, immunofluorescence assays with motor neurons labeled by SMI-32 were also performed, and the results indicated that the number of motor neurons in the spinal cord was lower in ALS model mice than in WT mice, and DLT treatment increased the number of motor neurons in the spinal cord in ALS model mice (Fig. 3d, e), verifying that DLT protected against motor neuronal loss in the spinal cords of ALS model mice.

DLT suppressed neuronal apoptosis in the spinal cords of ALS model mice

Given that neuronal apoptosis is the main reason for neuronal loss in the spinal cord of ALS [10], a TUNEL staining assay was performed. As shown in Fig. 3g, h, there were many TUNEL-positive cells in the spinal cords of ALS model mice, and treatment with DLT reduced the number of TUNEL-positive cells, implying that DLT suppressed spinal neuron apoptosis in ALS model mice.

Notably, AAV9-si-5Htr2a injection ameliorated motor neuronal damage in the spinal cords of ALS model mice, and DLT treatment did not exert the abovementioned ameliorative effects in AAV9-si-5Htr2a-injected ALS model mice (Fig. 3a, c, d, f, g, i).

Thus, the results demonstrated that DLT protected against spinal motor neuronal damage in ALS model mice through 5HTR_2A.

DLT promoted autophagy to clear spinal hSOD1^G93A proteins through the 5HTR_2A/cAMP/AMPK pathway in ALS model mice

DLT cleared spinal hSOD1^G93A proteins in ALS model mice through 5HTR_2A

Given that hSOD1^G93A protein aggregation is believed to be a major cause of ALS [42] and DLT was previously reported to reduce Aβ aggregation in the brains of APP/PS1 mice [24], we investigated whether DLT could also reduce spinal hSOD1^G93A protein aggregation in ALS model mice.

The Western blot results (Fig. 4a, b) indicated that the spinal hSOD1^G93A protein was highly expressed in ALS model mice, and DLT treatment reduced this protein level in ALS model mice. Notably, AAV9-si-5Htr2a injection reduced spinal SOD1^G93A protein levels in ALS model mice (Fig. 4a, c), but DLT treatment had no impact on SOD1G93A protein levels in AAV9-si-5Htr2a-injected ALS model mice (Fig. 4a, c). These results thereby demonstrated that DLT reduced spinal hSOD1^G93A protein levels in ALS model mice through 5HTR_2A.

Fig. 4. DLT promoted autophagy to clear spinal hSOD1^G93A protein in ALS mice.

a Western blot assay and (b, c) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection reduced the spinal hSOD1^G93A protein in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal hSOD1^G93A protein in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). d Immunofluorescence assay and (e, f) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection increased the protein level of spinal LC3 II in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal LC3 II in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). Scale bar: 10 µm. g Fluorescence assay and (h) its quantification results indicated that DLT cleared hSOD1^G93A protein level and co-treatment of autophagy inhibitor either 3-MA or CQ abolished DLT-mediated hSOD1^G93A protein clearance in NSC34 cells. Scale bar: 100 µm. All values were presented as mean ± SEM. ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

DLT promoted autophagy to clear spinal hSOD1^G93A proteins in ALS model mice through 5HTR_2A

In our previous work, DLT was reported to promote autophagy in the brains of APP/PS1 mice [24]. Here, the immunofluorescence assay results (Fig. 4d, e) indicated that spinal LC3II (autophagy marker protein) [43] levels were decreased in ALS model mice but increased in DLT-treated ALS model mice, indicating that the spinal autophagy level was reduced in ALS model mice and that DLT treatment promoted spinal autophagy. Notably, AAV9-si-5Htr2a injection promoted spinal autophagy in ALS model mice (Fig. 4d, f), and DLT treatment had no impact on spinal autophagy in AAV9-si-5Htr2a-injected ALS model mice (Fig. 4d, f). These results thus indicated that DLT promoted autophagy in ALS model mice through 5HTR_2A.

Furthermore, cell-based assays were carried out to verify that DLT reduced hSOD1^G93A protein levels in hSOD1^G93A-NSC34 cells by promoting autophagy. In the assay, NSC-34 cells stably overexpressing hSOD1^G93A protein and fused with GFP (hSOD1^G93A-NSC34 cells) were used. As shown in Fig. 3g, h, there was a large number of green fluorescent puncta in hSOD1^G93A-NSC34 cells, indicative of the overexpression of hSOD1^G93A protein, and DLT treatment reduced the number of green fluorescent puncta in the cells. Notably, cotreatment with the autophagy inhibitors 3-MA (inhibiting the generation of autophagosomes) [13] or chloroquine (CQ, blocking the fusion of autophagosomes and lysosomes) [44] abolished the ability of DLT to reduce hSOD1^G93A levels (Fig. 4g, h), which thus demonstrated that DLT reduced hSOD1^G93A protein levels by promoting autophagy in NSC34 cells.

The 5HTR_2A/cAMP/AMPK pathway was required for DLT-promoted spinal autophagy in ALS model mice

Since the 5HTR_2A/cAMP/SIRT1 pathway was found to be involved in the promotion of autophagy by DLT in the brains of APP/PS1 mice [24], we next determined whether this signaling pathway was also responsible for the DLT-promoted spinal autophagy in ALS model mice.

The immunofluorescence assay results indicated that the cAMP levels in both astrocytes (labeled by GFAP, Fig. 5a, b) and motor neurons (labeled by SMI-32 Fig. 5d, e) declined in the spinal cords of ALS model mice, and DLT treatment effectively reversed this cAMP level decline. The Western blot results (Fig. S4a, b) indicated that DLT had no effect on spinal SIRT1 protein levels in the spinal cords of ALS model mice. These results thus implied that SIRT1 signaling was not involved in the DLT-mediated autophagy process in the spinal cords of ALS model mice.

Fig. 5. DLT promoted autophagy to clear spinal hSOD1^G93A protein through 5HTR_2A/cAMP/AMPK pathway in ALS mice.

a Immunofluorescence assay and (b, c) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection increased spinal astrocytic cAMP level in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal astrocytic cAMP level in AAV9-si-5HTR_2A injected ALS mice (ALS-KD + DLT). Scale bar: 10 µm. d Immunofluorescence assay and (e and f) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection increased the cAMP level in the spinal motor neurons of ALS mice, and DLT treatment (DLT-20) had no impacts on the cAMP level in spinal motor neurons of AAV9-si-5HTR_2A injected ALS mice (ALS-KD + DLT). Scale bar: 10 µm. g Western blot and (h, i) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection increased spinal p-AMPK protein level in ALS mice, and DLT treatment (DLT-20) had no impacts on p-AMPK protein level in AAV9-si-5HTR_2A injected ALS mice (ALS-KD + DLT). j Immunofluorescence assay result indicated that DLT increased cAMP level in NSC-34 cells and co-treatment of TCB2 effectively deprived DLT of the regulatory effects on cAMP level. Scale bar: 50 µm. k Western blot assay results indicated that DLT increased protein level of p-AMPK in NSC-34 cells and co-treatment of TCB2 effectively deprived DLT of the regulatory effects on p-AMPK level. l Autophagic flow assay and (m) its quantification results indicated that co-treatment of Compound C abolished DLT-mediated autophagy activation in NSC34 cells. Scale bar: 150 µm. All values were presented as mean ± SEM. ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. **P < 0.01, ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

Considering that AMPK, as an autophagy regulator, can be activated by cAMP [45] and our previous work also determined that DLT could increase cAMP levels by antagonizing 5HTR_2A [24], we next investigated whether DLT might regulate AMPK in the spinal cords of ALS model mice. As expected, the Western blot results indicated that DLT increased spinal p-AMPK protein levels in ALS model mice (Fig. 5g, h). Thus, the results demonstrated that DLT activated the cAMP/AMPK pathway in the spinal cords of ALS model mice.

Notably, AAV9-si-5Htr2a activated the levels of cAMP (Fig. 5a, c, d, f) and p-AMPK (Fig. 5g, i) in the spinal cords of ALS model mice, and DLT treatment had no impact on cAMP/AMPK signaling in AAV9-si-5Htr2a-injected ALS model mice (Fig. 5a, c, d, f, g, i). These results thus indicated that DLT activated cAMP/AMPK signaling in ALS model mice through 5HTR_2A.

To verify the ability of DLT to regulate the 5HTR_2A/cAMP/AMPK pathway, the 5HTR_2A agonist TCB2 [46] was applied in the following in vitro assays. As shown in Fig. 5j, k (quantitative results in Fig. S4c, d), the immunofluorescence and Western blot results indicated that DLT increased the levels of cAMP and p-AMPK in NSC-34 cells, while the 5HTR_2A agonist TCB2 (1, 2 and 5 μM) effectively deprived DLT of its regulatory effects on cAMP and p-AMPK levels in NSC-34 cells and primary astrocytes. All results thus demonstrated the ability of DLT to regulate the 5HTR_2A/cAMP/AMPK pathway.

Furthermore, cell-based assays using the motor neuronal cell line NSC34 were also carried out to verify that cAMP/AMPK signaling is required for DLT-promoted autophagy. In the assay, an overexpression vector encoding mTagRFP-mWasabi-LC3 was applied to monitor autophagy. Briefly, the green and red puncta indicated mWasabi-LC3 and autolysosomes, respectively, and merged puncta (yellow) of green and red fluorescent signals indicated autophagosomes. As shown in Fig. S4e, f, DLT treatment increased the number of autophagosomes (red +; green +) and autolysosomes (red +) in the cells, indicating that DLT promoted autophagy in NSC34 cells. Notably, cotreatment with the AMPK inhibitor Compound C [47] abolished the ability of DLT to promote autophagy in NSC34 cells (Fig. 5l, m), thereby demonstrating that cAMP/AMPK signaling is required for DLT to promote autophagy.

Taken together, DLT promoted autophagy to clear the spinal hSOD1^G93A protein through the 5HTR_2A/cAMP/AMPK pathway in ALS model mice.

DLT suppressed spinal oxidative stress through 5HTR_2A/cAMP/AMPK/Nrf2-HO-1/NQO-1 pathway in ALS mice

Considering that aggregated hSOD1^G93A protein induces neuronal apoptosis by mediating oxidative stress (OS) [32], we investigated the potential of DLT in suppressing spinal OS in ALS model mice.

DLT suppressed spinal OS through 5HTR_2A/Nrf2-HO-1/NQO-1 signaling

To evaluate the capability of DLT in suppressing OS, the level of the lipid peroxidation product malondialdehyde (MDA [48]) a marker of OS in mouse plasma, was first determined by a commercial kit. The results indicated that the MDA level was increased in ALS model mice and decreased in DLT-treated ALS model mice (Fig. 6a), indicating the inhibition of OS by DLT.

Fig. 6. DLT suppressed spinal oxidative stress through 5HTR_2A/cAMP/AMPK/Nrf2-HO-1/NQO-1 pathway in ALS mice.

a, b The MDA assay results indicated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection reduced spinal MDA level in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal MDA level in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). c–h The RT-PCR results indicated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection upregulated the mRNA levels of Nrf2, HO-1 and NQO-1 in the spinal cord of ALS mice, and DLT treatment (DLT-20) had no further impacts on any of these anti-oxidation factors in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). i The ROS level assay results indicated that co-treatment of Compound C abolished the anti-oxidation of DLT in H₂O₂-treated NSC34 cells. j The DHE staining assay and (k) its quantification results indicated that co-treatment of Compound C abolished the anti-oxidation of DLT in H₂O₂-treated NSC34 cells. Scale bar: 150 µm. l–n RT-PCR assay results indicated that co-treatment of Compound C abolished the DLT mediated-upregulation of anti-oxidant factors (Nrf2, HO-1 and NQO-1) in H₂O₂-treated NSC34 cells. All values were presented as mean ± SEM. ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. *P < 0.05, ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

Next, given the important role of Nrf2-HO-1/NQO-1 signaling in OS [49], we investigated the potential effects of DLT on this signaling pathway in the spinal cords of ALS model mice. The RT‒PCR results indicated that the mRNA levels of spinal Nrf2 (Fig. 6c), HO-1 (Fig. 6e) and NQO-1 (Fig. 6g) in ALS model mice were higher than those in the WT group, reflecting the protective mechanism of the organism against the increased OS in ALS model mice. As indicated in Fig. 6c, e, g, DLT treatment activated spinal Nrf2-HO-1/NQO-1 signaling in ALS model mice, demonstrating the antioxidant effect of DLT. Notably, AAV9-si-5Htr2a suppressed OS (Fig. 6b) and activated spinal Nrf2-HO-1/NQO-1 signaling (Fig. 6d, f, h) in ALS model mice, while DLT exerted no effects on OS or Nrf2-HO-1/NQO-1 signaling in AAV9-si-5Hrt2a-injected ALS model mice (Fig. 6b, d, f, h).

All results thus demonstrated that DLT suppressed spinal OS through 5HTR_2A/Nrf2-HO-1/NQO-1 signaling in ALS model mice.

DLT suppressed H₂O₂-induced OS in NSC34 cells

Furthermore, assays with the motor neuronal cell line NSC34 were also performed to verify the antioxidant effect of DLT in vitro.

As shown in Fig. S5a, b, H₂O₂ treatment (300 μM) reduced cell viability and increased ROS levels in the cells, and DLT (2 and 5 μM) treatment effectively antagonized the H₂O₂-induced decline in cell viability and increase in ROS levels. Additionally, peroxide levels were also detected by an MDA detection kit and DHE staining in NSC34 cells, and the results demonstrated that DLT (2, 5 μM) abrogated the H₂O₂-induced increase in peroxide levels in both assays (Fig. S5c–e). Moreover, the regulatory effects of DLT on the antioxidant factors Nrf2, NQO-1 and HO-1 in cells were also determined. As shown in Fig. S5f–h, H₂O₂ treatment (300 μM) reduced the mRNA levels of Nrf2 (Fig. S5f), HO-1 (Fig. S5g) and NQO-1 (Fig. S5h), while DLT (2 and 5 μM) effectively increased the mRNA levels of these antioxidant factors in NSC34 cells.

Thus, the results indicated that DLT suppressed neuronal OS in the spinal cords of ALS model mice through 5HTR_2A.

cAMP/AMPK signaling was needed for the DLT to suppress spinal OS

Since we have determined that DLT exerts if effects by regulating spinal cAMP/AMPK signaling, assays were also performed to investigate the role of this signaling in the antioxidant effects of DLT.

As shown in Fig. 6i–n, cotreatment with Compound C (10 μM) abolished the decrease in ROS levels by DLT (ROS, Fig. 6i; DHE, Fig. 6j, k) and antioxidant factor upregulation (Nrf2, Fig. 6l; HO-1, Fig. 6m; NQO-1, Fig. 6n) in NSC34 cells, thus indicating that DLT suppressed H₂O₂-mediated OS in NSC34 cells through AMPK signaling.

Together, these results demonstrated that DLT suppressed spinal OS through the 5HTR_2A/cAMP/AMPK/Nrf2-HO-1/NQO-1 pathway in ALS model mice.

DLT suppressed spinal astrocytic activation through the 5HTR_2A/cAMP/AMPK pathway in ALS model mice

DLT suppressed spinal astrocytic gliosis in ALS model mice through 5HTR_2A

As astrocytic gliosis is a potent marker of astrocytic activation in neuroinflammation [33], the effect of DLT on spinal astrocytic gliosis in ALS model mice was first assessed by an immunofluorescence assay detecting astrocytes (GFAP). As shown in Fig. 7a, b, there were many active spinal astrocytes in ALS model mice compared with WT mice, and DLT treatment reduced the number of active astrocytes in ALS model mice. Additionally, AAV9-si-5Htr2a injection was found to suppress astrocytic gliosis (Fig. 7c, d) in the spinal cords of ALS model mice, while DLT treatment had no impact on astrocytic gliosis in AAV9-si-5Htr2a-injected ALS model mice. Thus, these results demonstrated that DLT suppressed spinal astrocytic activation in ALS model mice through 5HTR_2A.

Fig. 7. DLT suppressed spinal astrocytic activation through 5HTR_2A/cAMP/AMPK pathway in ALS mice.

a Immunofluorescence assay and (b, c) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection suppressed spinal astrocytic gliosis in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal astrocytic gliosis in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). Scale bar: 50 µm. d Immunofluorescence assay result indicated that DLT increased cAMP level in astrocytes and co-treatment of TCB2 effectively deprived DLT of its regulatory effects on cAMP level. Scale bar: 50 µm. e Western blot assay results indicated that DLT increased protein level of p-AMPK in astrocytes and co-treatment of TCB2 effectively deprived DLT of the regulatory effects on p-AMPK level. f Immunofluorescence assay and (g) its quantification results results demonstrated that DLT (2 and 5 μM) suppressed astrocytic activation in LPS/ATP-treated astrocytes. Scale bar: 50 µm. h Immunofluorescence assay and (i) its quantification results demonstrated that either DLT or si-5Htr2a treatment suppressed astrocytic activation in LPS/ATP-treated astrocytes, and si-5Htr2a treatment deprived DLT of its regulating activity against astrocytic activation in LPS/ATP-treated astrocytes. Scale bar: 50 µm. j Immunofluorescence assay and (k) its quantification results demonstrated that co-treatment of Compound C abolished DLT-mediated astrocytic activation suppression in LPS/ATP-treated astrocytes. Scale bar: 50 µm. All values were presented as mean ± SEM. ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

DLT suppressed spinal astrocytic activation through the 5HTR_2A/cAMP/AMPK pathway in primary astrocytes

Furthermore, related assays were performed to verify the role of the 5HTR_2A/cAMP/AMPK pathway in the DLT-induced suppression of astrocytic activation. First, the 5HTR_2A agonist TCB2 was applied in in vitro assays using primary astrocytes to confirm the ability of DLT to regulate the 5HTR_2A/cAMP/AMPK pathway in astrocytes. As shown in Fig. 7d, e (quantitative results in Fig. S6a, b), DLT increased the levels of cAMP and p-AMPK in primary astrocytes, while the 5HTR_2A agonist TCB2 (1, 2 and 5 μM) effectively abolished the ability of DLT to regulate cAMP and p-AMPK levels, demonstrating that DLT regulates the 5HTR_2A/cAMP/AMPK pathway in astrocytes through 5HTR_2A.

Then, si-5Htr2a and the AMPK inhibitor Compound C were applied in the following assays. In the assay, ATP (3 mM)/LPS (100 μg/mL) was used to activate astrocytes according to a published approach [18]. As shown in Fig. 7f, g, ATP/LPS treatment increased GFAP fluorescence intensity, and DLT (2, 5 μM) effectively reduced the fluorescence intensity of GFAP in ATP/LPS-treated primary astrocytes, indicating that DLT suppressed astrocytic activation in primary astrocytes. Notably, treatment with si-5Htr2a (Fig. 7h, i) or Compound C (10 μM, Fig. 7j, k) deprived DLT of its ability to suppress astrocytic activation.

The results thus indicated that DLT suppressed spinal astrocytic activation through the 5HTR_2A/cAMP/AMPK pathway in ALS model mice.

DLT suppressed spinal astrocytic NLRP3 inflammasome activation through the 5HTR_2A/cAMP/AMPK/NF-κB/NLRP3 pathway in ALS model mice

DLT suppressed spinal astrocytic NLRP3 inflammasome activation in ALS model mice through 5HTR_2A

Given that SOD1^G93A protein aggregation, as a key to ALS progression, activates the NLRP3 inflammasome to release the inflammatory cytokine IL-1β, leading to motor neuronal injury [15], we investigated the potential of DLT in regulating spinal NLRP3 inflammasome activation in ALS model mice.

Western blot (Fig. 8a, b) and RT‒PCR (Fig. S6c, e, g) results indicated that the protein and mRNA levels of the spinal NLRP3 inflammasome-related proteins NLRP3 (Fig. 8a, b and Fig. S6c), caspase-1p20 (Fig. 8a, b), ASC (Fig. 8a, b and Fig. S6e) and IL-1β (Fig. 8a, b and Fig. S6g) were all increased in ALS model mice and decreased in DLT-treated ALS model mice, demonstrating that DLT repressed spinal NLRP3 inflammasome activation in ALS model mice.

Fig. 8. DLT suppressed spinal astrocytic NLRP3 inflammasome activation through 5HTR_2A/cAMP/AMPK/NF-κB pathway in ALS mice.

a Western blot results and (b, c) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection reduced the protein levels of NLRP3, caspase-1 p20, ASC and IL-1β in the spinal cord of ALS mice, and DLT treatment (DLT-20) had no impacts on NLRP3 inflammasome related-proteins in the spinal cord of AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). (d) Immunofluorescence assay and (e, f) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection reduced the spinal astrocytic NLRP3 level in the ALS mice, and DLT treatment (DLT-20) had no impacts on spinal astrocytic NLRP3 level in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). Scale bar: 20 µm. g Immunofluorescence assay and (h, i) its quantification results demonstrated that either DLT treatment or AAV9-si-5Htr2a (ALS-KD) injection suppressed spinal astrocytic NF-κB nuclear translocation in ALS mice, and DLT treatment (DLT-20) had no impacts on spinal astrocytic NF-κB nuclear translocation in AAV9-si-5Htr2a injected ALS mice (ALS-KD + DLT). Scale bar: 20 µm. j Immunofluorescence assay and (k) its quantification results demonstrated that co-treatment of Compound C abolished DLT-mediated NLRP3 suppression in LPS/ATP-treated astrocytes. Scale bar: 50 µm. l, m RT-PCR assay results demonstrated that co-treatment of Compound C abolished DLT-mediated NLRP3 and IL-1β suppression in LPS/ATP-treated astrocytes. n Immunofluorescence assay and (o) its quantification results demonstrated that co-treatment of Compound C abolished DLT-mediated NF-κB nuclear translocation suppression in LPS/ATP-treated astrocytes. Scale bar: 50 µm. All values were presented as mean ± SEM. ^##P < 0.01, ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. *P < 0.05, **P < 0.01, ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

Next, the cellular localization of the NLRP3 inflammasome was investigated by immunofluorescence assay to identify which glial cells were responsible for NLRP3 inflammasome activation in the spinal cords of ALS model mice. As indicated in Fig. 8d, e, NLRP3 was mainly located in spinal astrocytes accompanied by astrocytic activation in ALS model mice and marginally located in microglia (Fig. S6i). These results demonstrated that astrocytes were the major cell population responsible for mediating neuroinflammation in the spinal cords of ALS model mice. As indicated in Fig. 8d, e, DLT treatment reduced the number of NLRP3-positive astrocytes in the spinal cords of ALS model mice, suggesting that DLT suppressed spinal astrocytic NLRP3 inflammasome activation in ALS model mice.

Notably, AAV9-si-5Htr2a injection suppressed spinal astrocytic NLRP3 inflammasome activation (Fig. 8a, c, d, f; Fig. S6d, f, h) in ALS model mice, while DLT treatment had no impact on the NLRP3 inflammasome in AAV9-si-5Htr2a-injected ALS model mice (Fig. 8a, c, d, f; Fig. S6d, f, h). Thus, the results demonstrated that DLT suppressed spinal astrocytic NLRP3 inflammasome activation in ALS model mice through 5HTR_2A.

DLT suppressed spinal astrocytic NF-κB nuclear translocation in ALS model mice through 5HTR_2A

Given that NF-κB, as a key nuclear transcription factor, regulates the expression of NLRP3 and ASC, we determined whether DLT suppressed NLRP3 inflammasome activation through NF-κB signaling in spinal astrocytes. As shown in Fig. 8g, h, NF-κB levels in the nucleus were higher in ALS model mice than in WT mice, and DLT treatment suppressed this level in ALS model mice, indicating that DLT suppressed spinal NF-κB activation in ALS model mice.

Notably, AAV9-si-5Htr2a injection suppressed spinal astrocytic NF-κB nuclear translocation (Fig. 8g, h) in ALS model mice, and DLT treatment had no impact on the level of nuclear NF-κB in AAV9-si-5Htr2a-injected ALS model mice. Thus, these results demonstrated that DLT suppressed spinal astrocytic NF-κB activation in ALS model mice through 5HTR_2A.

The 5HTR_2A/cAMP/AMPK pathway was required for the DLT-mediated suppression of the NLRP3 inflammasome in primary astrocytes

Furthermore, an in vitro assay using primary astrocytes was performed to further confirm the antagonistic effect of DLT on astrocytic NLRP3 inflammasome activation. The NLRP3 inflammasome is achieved by two sequential steps (termed priming and assembly) [17] and LPS as the priming signal and ATP as the secondary signal results in potassium efflux, ROS generation or lysosome rupture, and ultimately the assembly of the NLRP3 inflammasome [50]. Therefore, ATP (3 mM) and LPS (100 ng/mL) was used in the assay. Immunofluorescence (Fig. S7a, b), Western blot (Fig. S7c–e) and RT-PCR (Fig. S7f, g) assay results showed that ATP/LPS treatment increased the number of NLPR3-positive puncta, protein levels of NLRP3 and Caspase1 p20 and mRNA levels of NLRP3 and IL-1β in primary astrocytes, while DLT effectively reduced NLRP3 inflammasome activation in ATP/LPS-treated primary astrocytes. Additionally, DLT also suppressed ATP/LPS-induced NF-κB nuclear translocation in primary astrocytes (Fig. S7h, i). Notably, the immunofluorescence assay results showed that si-5Htr2a treatment abolished the ability of DLT to regulate NLRP3 expression (Fig. S7j, k) and NF-κB nuclear translocation in primary astrocytes (Fig. S7l, m). Moreover, as shown in Fig. 8j–o, cotreatment with Compound C (10 μM) effectively abolished the DLT-induced suppression of NLRP3 inflammasome activation (protein level, NLRP3, Fig. 8j, k; mRNA levels, NLRP3 and IL-1β, Fig. 8l, m) and NF-κB translocation (Fig. 8n, o), indicating that DLT suppressed astrocytic neuroinflammation through the 5HTR_2A/cAMP/AMPK pathway.

Together, DLT suppressed spinal astrocytic neuroinflammation through the 5HTR_2A/cAMP/AMPK/NF-κB/NLRP3 pathway in ALS model mice.

DLT suppressed the crosstalk between spinal astrocytic neuroinflammation and alleviated neuronal damage through 5HTR_2A

Since the crosstalk between spinal astrocytes and neurons is believed to be closely associated with ALS progression [16, 51], we also investigated the effects of DLT on astrocyte-mediated neuroinflammation and motor neuronal damage by a conditioned medium-based assay. Briefly, conditioned medium was obtained from spinal astrocytes treated with DMSO, LPS/ATP, LPS/ATP + DLT, si-5Htr2a, si-5Htr2a + LPS/ATP or si-5Htr2a + LPS/ATP + DLT and then applied to cultured motor neuronal NSC34 cells (Fig. 9a). As far as the compositions of the conditioned medium are concerned, the MTT assay results showed that DLT ameliorated the LPS/ATP-induced cell viability declines in NSC34 cells (Fig. 9b), and si-5-Htr2a abolished the ability of DLT to ameliorate neuronal damage (Fig. 9b). Additionally, the TUNEL staining assay results also indicated that DLT decreased the apoptosis of NSC34 cells treated with conditioned medium (Fig. 9c, d), and si-5Htr2a treatment abolished the ability of DLT to decrease NSC34 cell apoptosis (Fig. 9b–d).

Fig. 9. DLT repressed the crosstalk between spinal astrocytic neuroinflammation and neuronal damage through 5HTR_2A.

a Schematic diagram of conditioned medium experiment (astrocytes-neurons). b MTT results indicated that treatment of DLT (5 μM) in the conditioned medium from the ATP/LPS-induced inflammatory astrocytes antagonized the neuronal cell viability declines through 5HTR_2A in NSC34 cells. c TUNEL staining and (d) its quantification results demonstrated that treatment of DLT (5 μM) in the conditioned medium from the ATP/LPS-induced inflammatory astrocytes antagonized the neuronal apoptosis through 5HTR_2A. Scale bar: 150 μm. e Schematic diagram of conditioned medium experiment (neurons-astrocytes). f Immunofluorescence and (g) its quantification results demonstrated that treatment of DLT (5 μM) in the conditioned medium from the damaged neurons antagonized the astrocytic activation and NLRP3 upregulation through 5HTR_2A. Scale bar: 50 μm. h Immunofluorescence and (i) its quantification results demonstrated that treatment of DLT (5 μM) in the conditioned medium from the damaged neurons antagonized the astrocytic NF-κB nuclear translocation through 5HTR_2A. Scale bar: 50 µm. j Fluorescence assay and (k) its quantification results demonstrated that si-H1R had no effect on the regulation of DLT against mutated SOD1 in NSC-34 cells. Scale bar: 100 µm. l ROS level assay results demonstrated that si-H1R had no effect on the regulation of DLT against mutated ROS level in H₂O₂-treated NSC-34 cells. m Immunofluorescence assay and (n) its quantification results demonstrated that si-H1R had no effects on the regulation of DLT against NLRP3 level in LPS/ATP-treated astrocytes. Scale bar: 50 µm. All values were presented as mean ± SEM. ^###P < 0.001 compared with WT or WT-NC mice by one-way ANOVA test. ***P < 0.001 compared with ALS or ALS-NC mice by one-way ANOVA test.

Next, we investigated the relationship between motor neuronal damage and spinal astrocytic neuroinflammation. In the assay, conditioned medium from damaged motor neurons was used to culture spinal primary astrocytes (Fig. 9e). The immunofluorescence assay results indicated that DLT inhibited astrocytic activation (marked by GFAP), NLRP3 upregulation (Fig. 9f, g) and NF-κB nuclear translocation (Fig. 9h, i) induced by the conditioned medium from damaged motor neurons in NSC34 cells, and si-5Htr2a abolished these DLT-mediated effects (Fig. 9f–i).

Together, these results indicated that DLT suppressed the crosstalk between spinal astrocytic neuroinflammation and alleviated motor neuronal damage through 5HTR_2A.

DLT suppressed ALS-like pathology in an H1R-independent manner

Considering that DLT, as an H1R inhibitor, is used to treat allergic diseases in clinical practice, in vitro assays were performed containing a negative control (H1R was knocked down by si-h1r, Fig. S7n, o) to investigate whether the inhibition of H1R by DLT was involved in its anti-ALS pathology effects. As shown in Fig. 6j–n, H1R knockdown failed to affect the regulation of mutated SOD1 clearance (Fig. 9j, k), oxidative stress (Fig. 9l) or inflammatory events (Fig. 9m, n) by DLT, thus suggesting that DLT ameliorated ALS-like pathology in an H1R-independent manner.

Discussion

5HTR_2A is widely expressed in the CNS and participates in various biological events, including endogenous transmitter secretion and sensory signal transduction [19]. It was reported that 5HTR_2A activation is often accompanied by astrocytic activation [22]. Herein, we determined that astrocyte activation is increased and 5HTR_2A expression is upregulated in the spinal cords of ALS model mice and that DLT, as a 5HTR_2A antagonist, suppressed astrocytic activation and alleviated the loss of motor neurons in ALS model mice. Our results strongly support that spinal 5HTR_2A-mediated astrocytic activation might play an important role in ALS progression. Notably, it was difficult to clearly define the beneficial effects of DLT on ALS model mice by focusing on motor neurons or astrocytes due to the simultaneous alleviation of the astrocytic (neuroinflammation) and motor neurogenic (mutated SOD1 level, oxidative stress) pathology by DLT and the complexity of the pathogenesis of ALS. However, it has been reported that neuroinflammation, as an early pathological feature in the spinal cord of ALS, mainly drives the loss of motor neurons [52] and the fact that astrocytes expressing mutated SOD1 are prone to enhanced activation has suggested a potential role for astrocytes in controlling microglial activation and neuroinflammation [53]. Moreover, astrocytes are more susceptible to the accumulation of mutated SOD1 protein than motor neurons [51] and mutated SOD1-mediated astrocytic activation has been recognized as an upstream event of motor neuronal damage in ALS pathology [54]. Thus, we focused on investigating the effects of DLT on astrocyte-mediated cascade reactions in ALS pathology and tentatively suggested that the amelioration of astrocytic and motor neurogenic pathology might be synergistically responsible for the anti-ALS effect of DLT.

The mutant hSOD1^G93A protein is a key toxic protein implicated in ALS pathology, and clearance of this protein is considered an effective strategy for ALS treatment [55]. Autophagy, as an important intracellular clearance process, is responsible for the degradation of intracellular wastes, including damaged organelles and abnormal proteins [12]. In the current study, we determined that spinal autophagy was alleviated and that DLT promoted autophagy to reduce hSOD1^G93A protein levels in ALS model mice. The increased accumulation of toxic proteins such as the hSOD1^G93A protein due to autophagy disorder is an important feature of ALS [55]. These results have thus provided evidence that autophagy activators such as DLT may have therapeutic functions in ALS.

Neuroinflammation leads to neuronal damage in ALS, and astrocytes, the largest type of glial cell, are activated and proliferate in the brainstem and spinal cord of ALS patients and model mice [56, 57]. ALS-associated mutated genes such as SOD1^G93A are overexpressed in astrocytes, reflecting the role of neuroinflammation in ALS [51]. Inflammation-activated astrocytes damage the surrounding neurons by releasing inflammatory factors, and damaged neurons induce an inflammatory reaction in surrounding astrocytes, producing a vicious cycle that eventually exacerbates the pathology of ALS [16, 33]. Thus, the ability of DLT to suppress OS and astrocytic neuroinflammation in the spinal cords of ALS model mice strongly supports the potential of this agent in treating ALS.

Interestingly, in contrast to our previous report that DLT ameliorated AD-like pathology through the 5HTR_2A/cAMP/SIRT1 pathway [24], we found here that DLT improved ALS-like pathology through the 5HTR_2A/cAMP/AMPK pathway. We speculated that this inconsistency between SIRT1 and AMPK signaling might be caused by the different pathological processes between AD and ALS. In the process of aging, mitochondrial function is slowly degraded, accompanied by a reduction in antiaging factors such as SIRT1 [58]. Unlike the slow neurodegeneration mainly caused by aging in AD [59], the neurodegenerative progress in ALS pathology is comparatively faster and more severe, and the energy metabolism dysfunctions cannot be ameliorated [60], thus resulting in the disorder of several energy-related proteins, including AMPK. It is expected that our findings may help determine the right therapeutic strategy for ALS.

In our study, we found that either DLT treatment or AAV9-si-5Htr2a injection effectively alleviated muscular dysfunction in ALS model mice, although no differences were observed in 5HTR_2A levels between WT and ALS model mice, and AAV9-si-5Htr2a injection failed to regulate 5HTR_2A levels in the muscles of ALS model mice. The muscular dysfunctions of ALS model mice are mainly attributed to motor neuron loss, which results from disuse atrophy and the neurogenic nutritional deficiency of muscles [61, 62]. Based on the above results, we proposed that DLT might repair the motor dysfunctions and neurotrophic state of muscle tissues controlled by the CNS by ameliorating ALS-like pathology in the spinal cord, resulting in the activation of muscular movement-mediated AMPK [63] and the improvement of mitochondrial dysfunction.

Notably, DLT has been reported to be a third-generation antihistamine agent and is clinically used to treat antiallergic and chronic urticaria [37]. Although the amelioration by DLT on ALS-like pathology has been verified to be independent of H1R, the already known clinical information of DLT, such as metabolism, toxicity, and pharmacokinetics, may help greatly reduce the cost and time of the research and development for anti-ALS reagents based on DLT as an available “old” drug.

Finally, since respiratory muscle failure in ALS patients may lead to increased tracheal secretions that cannot be excreted [3], antihistamine treatment is clinically required to alleviate the respiratory allergic response [60]. In that case, DLT is obviously more capable than other antihistamines of both alleviating respiratory allergic reactions and ameliorating ALS pathology in patients based on our current findings, thus possibly relieving the economic burden on society and family.

Conclusion

In conclusion, we found that DLT, as a 5HTR_2A antagonist, effectively delayed the onset time of ALS-like symptoms, prolonged the life cycle and improved dyskinesia, gastrocnemius injury and motor neuronal loss in ALS model mice. Moreover, we determined that DLT suppressed oxidative stress, promoted autophagy to clear the hSOD1^G93A protein and inhibited astrocytic neuroinflammation in the spinal cords of ALS model mice through the 5HTR_2A/cAMP/AMPK pathway. Our work has highlighted that 5HTR_2A antagonism shows promise as a therapeutic strategy for ALS and that DLT as a 5HTR_2A antagonist shows potential in treating this disease.

Author contributions

XS, JYW and JL designed the study. XS reviewed the manuscript. JL, ZYJ, AXH, MZ, ZXL, FZ and LM performed the animal and cell experiments. JL, ZYJ and AXH analyzed and interpreted data. JL wrote the manuscript. JL, XS, JYW and HMJ are the guarantors of this work and, as such, have full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors approved the manuscript.

Data availability

The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests. All institutional and national guidelines for the care and use of laboratory animals were followed.

Supplementary information

The online version contains supplementary material available at 10.1038/s41401-023-01223-2.